AI training for employees works best inside the tools people use every day, says ivee’s Amelia Miller. Businesses are buying AI seats faster than staff learn to use them, so the return never arrives. Miller, ivee’s co-founder and CEO, showed the fix live inside Claude at aiblLIVE London 2026.
In the demo, ivee rewrote a vague board-paper prompt in place and explained each change. The improved prompt made Claude ask questions before writing, at a fraction of the cost. Here is what she showed.
All figures as stated on stage by Amelia Miller.
Because people do not know how to use them. Miller calls this AI adoption debt: tools move faster than organisations adopt them. If someone has never opened Copilot, that seat returns nothing. ivee’s bet is that training each employee closes the gap, more than strategy alone.
Buying the seats is the easy part, but changing how people work in their everyday is definitely the hard part.
Amelia Miller, Co-Founder and CEO, ivee
A weak prompt fixed before it was sent. Miller asked Claude for a board paper with almost no context, and got a long, unsourced document. ivee then rewrote the prompt. It added a persona, a request to confirm structure first, cited sources, a critical check and rules against AI writing tells.
It replaces long courses with help at the moment of use. Each improved prompt shows what changed and why, so people learn as they work. A prompt library lets them reuse and share good prompts. Success is measured by how people’s behaviour changes, beyond course completion.
Team stats and shadow AI. Leaders see prompts improved, estimated time and token savings, and which tools staff use. An automation finder flags repeated tasks to automate. A shadow AI view shows unsanctioned tools in use, which may breach data policy.
We don’t want people to just live on their LLMs like a glorified Google chatbot. We want them to be automating.
Amelia Miller, Co-Founder and CEO, ivee
Amelia Miller is co-founder and CEO of ivee, the AI training company she founded in 2024 with her sister Lydia, its CTO. She is also a fellow at the University of Cambridge, studying skills prediction in the labour market. She spoke on the Efficiency track at aiblLIVE London on 20 October 2026.
If your AI licences are live but your returns are not, this is the training model to look at. The full PDF goes deeper than this page, in Miller’s own words.
For L&D leads, HR directors, COOs and anyone responsible for AI rollout. Five pages, about a seven-minute read.
